harness

分类: 测试与安全 | 上传者: 3yesore | 下载: 0 | 版本: v1.0（最新）

应用线束工程使模块可人工智能维护。在创建新模块、改造现有模块、验证模块合规性或保持模块跨会话准备就绪时使用。

更新日志: Source: GitHub https://github.com/3yesore/Harness-Tool-minimal

目录结构

当前层级: tree/main/.openclaw_skill/

📁 profiles/
- 📄 default.rules.json 373 B
- 📄 python-service.rules.json 448 B
📁 references/
- 📄 changelog_template.md 356 B
- 📄 harness_spec.md 4.0 KB
- 📄 index_template.md 476 B
- 📄 spec_template.md 995 B
- 📄 validation_rules.md 1.6 KB
📁 scripts/
- 📄 apply_harness.py 9.5 KB
- 📄 init_harness_module.py 2.8 KB
- 📄 validate_harness.py 10.8 KB
📁 templates/
- 📄 CHANGELOG.md.template 347 B
- 📄 INDEX.md.template 458 B
- 📄 SPEC.md.template 995 B
📄 SKILL.md 2.7 KB

SKILL.md

---
name: harness
description: Apply Harness Engineering to make modules AI-maintainable. Use when creating new modules, retrofitting existing modules, validating module compliance, or keeping a module handoff-ready across sessions.
---

# Harness Engineering

Make modules AI-maintainable through standardized interfaces, clear indexing, and runnable verification.

## What This Skill Includes

- `scripts/init_harness_module.py`: create a new module skeleton
- `scripts/apply_harness.py`: retrofit an existing module with Harness docs and a smoke test
- `scripts/validate_harness.py`: validate docs, configs, and smoke-test execution
- `templates/`: INDEX, SPEC, and CHANGELOG templates
- `profiles/`: rule presets used by the scripts

## Installation

Copy the `.openclaw_skill` directory into your OpenClaw skills folder:

```powershell
New-Item -ItemType Directory -Force "$env:USERPROFILE\.openclaw\skills" | Out-Null
Copy-Item -Recurse -Force .openclaw_skill "$env:USERPROFILE\.openclaw\skills\harness"
```

After installation, the skill is available as `harness`.

## Core Principles

1. Index-first: every maintained module should have `INDEX.md`
2. Contract-driven: `SPEC.md` should define inputs, outputs, config, and errors
3. Minimal context: keep maintenance-critical details in a few files
4. Verifiable: every module should have a runnable smoke test
5. Handoff-ready: another maintainer should recover the module quickly from `INDEX.md`, `SPEC.md`, and the smoke test

## Quick Reference

```bash
# Initialize new module
python scripts/init_harness_module.py my_module

# Apply Harness to existing module
python scripts/apply_harness.py existing_module

# Validate module compliance
python scripts/validate_harness.py my_module

# Validate with a profile
python scripts/validate_harness.py my_module --profile python-service

# Run module tests
python my_module/tests/smoke.py
```

## AI Workflow

1. Read `INDEX.md` and `SPEC.md`
2. Read only the code files needed for the task
3. Make the change
4. Run the smoke test
5. Run `validate_harness.py`
6. Update `CHANGELOG.md` when behavior changes

## Profile Support

Profiles are bundled with the skill and are consumed by the scripts through `--profile`.

- `default.rules.json`
- `python-service.rules.json`

The profiles are intentionally minimal and mostly mirror the default behavior today.

## What You Get From Skill Only

If you only install this skill, you get the current Harness workflow, templates, profiles, and validation scripts packaged together.  
You do not get future repository-only changes unless the skill is updated again.

## References

- `references/index_template.md`
- `references/spec_template.md`
- `references/changelog_template.md`
- `references/harness_spec.md`
- `references/validation_rules.md`

登录后下载/点赞/收藏 ❤ 6 | ★ 0

harness

目录结构

SKILL.md

举报内容

提示